Did CNNIBN-CSDS apply the sampling error manipulation trick for their Chattisgarh projections?
“In one instance, pollster Yashwant Deshmukh of CVoter, told the News Express channel’s undercover reporter that while 3% was the standard margin of error, “at best, we can put it to 5%”.
The duo, Rajdeep Sardesai (who works in Mukesh Ambani controlled CNNIBN) and Dr. Sandeep Shastri (of Ford Foundation funded CSDS) is back publishing a national opinion poll, beginning on Monday. Those who want to watch pure entertainment can tune in and enjoy one free.
Last November, both these characters were in their elements in trying to project Raman Singh's BJP government attracting what they described as a pro-incumbency advantage. They made this claim repeatedly in their opinion polls. Their last November 2013 opinion poll for the state of Chattisgarh can be compared with the actual results as provided above.
What the data clearly establishes is that CSDS over-estimated the BJP by 4% and under-estimated the Congress by a whopping 8% to provide the BJP a winning margin by an even more whopping 14%. Compare this to what Yashwant Deshmukh of CVoter (caught on tape by #OperationPrimeMinister) attributed as the industry standard of permissible limits to raw data tweaks to either favour or disfavour a political party or both – plus/minus 5%.
Obviously this duo Rajdeep Sardesai-Sandeep Shastri managed to do even much better the industry standards of data manipulation that can lead us to conclude that either the CNNIBN-CSDS are an apology for opinion polls in terms of their skill & accuracy and/or this duo perhaps are the most unscrupulous within the entire industry!
Poor Yashwant Deshkukh ended up in the doghouse summarily dismissed as a fraud even when there are actually likes of Rajdeep Sardesai-Sandeep Shastri still at large. It must be pointed the only reason why CNNIBN-IBN was not part of the #OperationPrimeMinister sting was that they declined to meet the sting team ostensibly since they claimed they were overbooked for the season and unable to take up new business! Media sources however suggest they were tipped off on the sting.
Of course you can bet both Rajdeep Sardesai and Sandeep Shastry would dish out a galore of spin to favour the opinion poll industry and their own credentials. You can take for granted too that their Chattisgarh predictions will never be a mentioned.
What they will probably not tell you are the following:
a.What are the credentials of Sandeep Shastry in psephology? Is he a trained psephologist as his qualifications is basically in Political Science?
b.What is the track record of CSDS forecast since Sandeep Shastry replaced Yogendra Yadav?
c.Is the editorial policy of CNNIBN politically neutral or posses a pro-Modi bias as widely perceived?
d.How have Rajdeep’s and his wife’s (Sagarika Ghose), Editor and Dy Editor of CNNIBN respectively, tweets complaining of media freedom being under threat influenced the bias of these polls?
e.What is the margin of error of these polls and in terms of vote-seat projections what does it amount to?
f.How do they explain projections like Chattisgarh – repeat “errors” where they projected BJP leads in double digits while the gap was as close as 0.6%?
g.Will they get critics of opinion polls and those discredited by #OperationPrimeMinister like Yashwant Deskukh on their show while spinning in favour of such polls like theirs
h.What is the source of funding? What is the total budget? What proportion is accounted by fieldwork?
BJP's creating a wave through rigging opinion polls go bust!
Fraud polls.....a sinister BJP strategy? (Prashant Pandey in The Real Truth)
The Economic Times talks of a sting operation conducted by a TV channel called News Express which shows that
"
undercover reporters agreed to manipulate poll data". It adds "
Clips from the sting operation aired by the channel showed many pollsters agreeing to produce favourable numbers by leveraging the so-called margin of error, a statistical concept meant to indicate the quality of sampling and the accuracy to be expected from survey results".
The fact is that the manipulation goes way beyond playing with statistical errors. The methods, and intentions, are far more sinister.
But before that, lets look at who has been the beneficiary of these fraud polls.
One single party, the BJP. The BJP has been showing rising with every poll, creating the illusion of a wave. The timing of the wave was always suspicious. It rose with the appointment of Narendra Modi first as the poll campaign chief of his party, and later, and at a much faster pace, after he was made the PM nominee. If someone is paying off the field researcher to show a favorable result, who could it be? I think we are smart enough to figure that out!
The methods, like I said earlier, are far more sinister than merely "leveraging" some "sample of error". That's just the talk of a guilty man trying to drown his crime in a lot of mumbo-jumbo. Of course, all researches have errors, but a research is designed in a way that keeps error at an acceptable level. And depending on the design and the sample size, certain data cuts are not permited because the error would be too high. Error is central to any sampling; there is no surprise, nor possibility of an excuse, in error rates.
But there are other more devious ways used to manipulate results. A field researcher may simply "fudge" the questionnaire, filling exactly what he wants to fill without so much as bothering to ask the respondent for his/her views. Or changing the respondent's answers even after recording them on the questionnaire.
Or entering the wrong responses into the "system" so as to suit the sponsor. This is all too common because the field agency is under pressure to deliver "cheap". Well, respondents deliver real quick by filling the questionnaires all up at home, or resorting to the tricks mentioned before! In the process, they also make themselves richer by a fair bit.
Don't believe this? Consider this. A 20,000 sample size research should cost up to Rs 2-3 crores at current rates. If the fieldwork involves extensive travel into remote villages, the costs could increase beyond this. Which news channel has so much budget? Many of these news channels do 3-5 polls before an election. How can they afford so many? I'll tell you how. They get it done cheap! (For the official records, they say that the research costs are shared with a newspaper, but in reality, that would only halve the cost....not make it so affordable).
The second sinister reason is even more sinister (perhaps). The researcher goes to a home, finds it to be a supporter of an "opposing" party, and simply skips the home! Simple and damned effective! He then goes to a home where he finds a supporter of his devious sponsor, and finishes the interview there. Clearly, the results will make the sponsor happy!
A 3rd devious design is when information of the "starting point" of the fieldwork gets known to the sponsor in advance. The starting point is where, typically, a researcher begins his survey, and to eliminate any sampling bias, he follows a"right hand rule"to cover the first few houses he encounters. If this starting point is known in advance, the sponsor plugs his messages into those homes in advance, thus influencing the minds of those respondents.
This is why polls results are so different from reality. In 2004 and 2009, the Congress was shown to be the loser, just like it is being shown today. In both times, the BJP was expected to "romp home". The reality couldnt have been farther from the truth. The BJP was trounced, the Congress emerged stronger in 2009. Take the recent assembly polls. All pollsters got AAP wrong. Why? Because maybe their sponsor was not AAP! Simple....In all examples, the BJP is the biggest beneficiary of these polls. Not surprising it is the only party that is opposing the Election Commission's own view that opinion polls should be banned.
In today's ET, BJP spokesperson Prakash Javdekar has given a silly statement
"
We haven’t yet demanded a ban because these are just opinion polls, and people vote on their own considerations".
Really? And how do people form their "considerations"? ONLY and ONLY via media. Consider this. Most people think UPA2 is very corrupt. How do these people have this opinion? Did any of them personally read the CAG's report on 2G or coal? Did any of them do any "chai pe charcha" (that hyped-up smokescreen to justify the fraud poll results) to unearth nuggets of wisdom from scratch? No. They all read the papers and worse....watched the news (often called Horror Entertainment Channels!).
The BJP's game plan could be to influence the voters and make them vote for it. That is why they are creating this bogus fear of a "hung Parliament", and the need for a "decisive verdict". An otherwise moderate Hindu, worried about the country's economic problems but not supporting the BJP's polarizing politics, could be made to swing towards the BJP by creating the impression that it is "nearly there". A Muslim who would traditionally vote for the SP in UP would swing towards the BSP thinking "all" have shifted loyalties....helping divide the Muslim vote and the BJP.
Influencing public opinion is an attack on democracy. It is fooling the public. How can a party that does this be trusted? Already we've heard of noted journalists like Sagarika Ghose, Siddharth Varadarajan, Hartosh Singh Bal.....and dare I add the name of Tarun Tejpal to this list....being ousted from their jobs for political reasons (they are no Congress afficianados by the way).
The party that influences public opinion and gets after journalists won't think twice before muting out every opposing voice when it comes to power. Don't believe it? See how Muslims in Gujarat hardly have any voice left. They dont dare to even carry out a street morcha! When Haren Pandya started to croak, he was silenced by someone (His family accuses Modi; not proven though).
The real truthis that I have always said that opinion polls were fraudulent. It's now been proven by this sting operation. Fortunately, Times Now has said it wont use C-Voter again. But will they use someone else who might be as fraudulent? The problem is not with C-Voter. The problem is with the party that funds these pollsters....the party that is shown as the beneficiary?